基于多分辨率混合特征的色纺织物组织识别  

Colored spun fabric pattern recognition based on multi-resolution mixed characteristics

作　　者：王丹书 杨亚莉 袁理[1,2] WANG Danshu;YANG Yali;YUAN Li(School of Electronic and Electrical Engineering,Wuhan Textile University,Wuhan 430200,China;State Key Laboratory of New Textile Materials and Advanced Processing Technology Jointly Built by the Province and the Ministry,Wuhan Textile University,Wuhan 430200,China)

机构地区：[1]武汉纺织大学省部共建纺织新材料与先进加工技术国家重点实验室,武汉430200 [2]武汉纺织大学电子与电气工程学院,武汉430200

出　　处：《丝绸》2021年第11期27-32,共6页Journal of Silk

基　　金：湖北省自然科学基金项目(2014CFB754);湖北省教育厅科学技术研究计划青年人才项目(Q20141607);中国纺织工业联合会科技指导性项目(2018035,2014072)。

摘　　要：针对色纺织物组织识别过程中参数特征提取困难的问题,文章提出了一种基于多分辨率混合特征的织物组织结构自动识别算法。该算法利用小波函数对色纺织物图像进行多分辨率分解,并提取2DLBP纹理和全局颜色特构建混合特征向量;同时,采用BP网络和朴素贝叶斯理论构建层次化分类器,从而实现组织特征参数的分类与识别。实验结果表明,提取的织物组织特征不仅具备颜色与多层纹理表征能力,而且所构建的混合分类器也兼具强化特性。对180份具有不同染色纤维混配系数及捻系数的色纺织物组织图像进行识别,平均识别率约为97%,验证了方法的有效性与鲁棒性。To resolve the difficulty in extracting the characteristics during the process of colored spun fabric pattern recognition,an automatic recognition algorithm based on multi-resolution mixed characteristics was proposed.Multi-resolution decomposition was conducted in the algorithm by means of wavelet function and 2DLBP textures and global color features were extracted to establish a mixed characteristic vector.At the same time,a hierarchical hybrid classifier was established using BP network and naive Bayes theory to realize the classification and recognition of the structure characteristic parameters.The experimental results show that the extracted weave pattern characteristics have the capability of color and multilayer characterization,and the hybrid classifier also has the characteristics of enhancement.180 colored spun fabric pattern images with different mixed coefficients and twist coefficients were identified,with an average recognition rate of 97%,which proved the effectiveness and robustness of this method.

关 键 词：色纺织物 组织识别 小波分解 2DLBP 混合特征 

分 类 号：TS101.9[轻工技术与工程—纺织工程]